State-based control for organizationally situated agents

Local agent control for sophisticated agents situated in open, dynamic environments is a complex problem. Such agents are hindered by bounded rationality, limited resources, and imperfect knowledge of the world and the activities being performed by other agents. Openness and dynamism conspire against an agent's ability to schedule or plan far downstream temporally-leading to a requirement for efficient or soft real-time online control problem solving. To further complicate the problem, agents in open environments or large scale multi-agent systems must also be able to reason about their different relationships with other agents and the different organizational objectives associated with different organizations to which they belong, all the while addressing resource limitations in their activities. We view local agent control as an action-selection-sequencing problem where an agent has n candidate tasks and alternative ways to perform the tasks. Tasks have deadlines and other constraints as well as different performance properties, e.g., consuming different resources or producing results of varying quality. Agent control from this view is an optimization problem; the problem is to choose which tasks to perform and how to perform them where the appropriate choice is dependent on the agent's context, which includes its relationships with other agents, shared organizational goals, individual organizational goals, commitments made with other agents, and resource limitations.